De novo dominant ASXL3 mutations alter H2A deubiquitination and transcription in Bainbridge-Ropers syndrome

Abstract

De novo truncating mutations in Additional sex combs-like 3 (ASXL3) have been identified in individuals with Bainbridge-Ropers syndrome (BRS), characterized by failure to thrive, global developmental delay, feeding problems, hypotonia, dysmorphic features, profound speech delays and intellectual disability. We identified three novel de novo heterozygous truncating variants distributed across ASXL3, outside the original cluster of ASXL3 mutations previously described for BRS. Primary skin fibroblasts established from a BRS patient were used to investigate the functional impact of pathogenic variants. ASXL3 mRNA transcripts from the mutated allele are prone to nonsense-mediated decay, and expression of ASXL3 is reduced. We found that ASXL3 interacts with BAP1, a hydrolase that removes mono-ubiquitin from histone H2A lysine 119 (H2AK119Ub1) as a component of the Polycomb repressive deubiquitination (PR-DUB) complex. A significant increase in H2AK119Ub1 was observed in ASXL3 patient fibroblasts, highlighting an important functional role for ASXL3 in PR-DUB mediated deubiquitination. Transcriptomes of ASXL3 patient and control fibroblasts were compared to investigate the impact of chromatin changes on transcriptional regulation. Out of 564 significantly differentially expressed genes (DEGs) in ASXL3 patient fibroblasts, 52% were upregulated and 48% downregulated. DEGs were enriched in molecular processes impacting transcriptional regulation, development and proliferation, consistent with the features of BRS. This is the first single gene disorder linked to defects in deubiquitination of H2AK119Ub1 and suggests an important role for dynamic regulation of H2A mono-ubiquitination in transcriptional regulation and the pathophysiology of BRS.

Figure 1.

Characterization of pathogenic truncating ASXL3 variants. (A) Facial characteristic of proband I at age 36 months showing low-set ears, broad nasal bridge, sparse arched eyebrows, downslanting palpebral fissures and broad forehead with periorbital fullness. (B) Brain MRI of proband I showing cerebellar vermal hypoplasia (single star), corpus callosum hypoplasia (white arrows), increased subdural space (two red arrows) and decreased white matter volume. (C) Proband II has a small chin, downslanting palpebral fissures and borderline low-set ears. (D) Proband II showing cerebellar vermal hypoplasia (single star) and a shortened corpus callosum (white arrow). (E) Schematic illustration of the main functional units of ASXL3 and pathogenic variants. Novel ASXL3 variants are depicted by red arrows. Previously described truncating ASXL3 variants (orange arrows cluster in the MCR, orange box). Missense ASXL3 variants identified in individuals with Autism spectrum disorder (green stars). ASXL3 functional units are shown with corresponding amino acids. ASXL3 exons 11 and 12 are 1.9 and 3.7 kb in length, respectively, and comprise ∼5.6 kb of the ∼6.8 kb ASXL3 open reading frame.

Figure 2.

Characterization of truncating ASXL3 variants in patient fibroblasts. (A) Schematic of ASXL3 C-terminally V5/His tagged constructs and length in amino acids (aa), the region of ASXL3 recognized by the α-ASXL3 antibody (orange antibody), ASXL3 N-terminus that shares 89% homology with the BAP1-binding region of ASXL1 and 2 (gray bar), proband I truncating variant (red arrow). (B) Western blot of two biological replicates comparing endogenous full-length ASXL3 expression in patient ASXL3^+/fs and control ASXL3^+/+ fibroblasts (+/fs to +/+). Overexpressed fl-ASXL3-V5/His runs at ∼280 kDa, corresponding to the highest molecular weight band from fibroblast cell lysates recognized by the α-ASXL3 antibody. fl-ASXL3-V5/His is recognized by α-ASXL3 and α-V5 antibodies. β-Actin was used as a loading control. (C) Quantification of endogenous full-length ASXL3 protein in ASXL3^+/fs and control ASXL3^+/+ fibroblasts relative to expression of β-actin, n = 3 (Error bars denote SD; *P = 0.04). (D) Overexpressed tr-ASXL3-V5/His is recognized by α-ASXL3 and α-V5 antibodies. A band of comparable molecular weight is not detected in ASXL3^+/fs and control ASXL3^+/+ cell lysate. (E) qRT–PCR analysis of ASXL3 transcripts from ASXL3^+/fs and control ASXL3^+/+ fibroblasts cDNA relative to GAPDH, n = 3 (Error bars denote SD; **P = 0.005).

Figure 3.

ASXL3 interacts with BAP1 and truncating mutations alter histone H2A ubiquitination. (A) BAP1-Flag and tr-ASXL3-V5/His were individually and co-overexpressed in 293 T. The N-terminal fragment of ASXL3 is post-transcriptionally modified when co-overexpressed together with BAP1-Flag and is seen as a doublet in the corresponding lane. Reciprocal co-immunoprecipitations were performed with α-V5 and α-flag antibodies. BAP1-Flag preferentially binds the post-transcriptionally modified tr-ASXL3-V5/His fragment. (B) H2AK119Ub1 and H3K27Me3 acid extracted histones levels compared between ASXL3^+/+ and ASXL3^+/fs fibroblasts. H3 used as loading control. A 5-fold increase in H2AK119Ub1 was quantified in ASXL3^+/fs fibroblasts compared with H3 levels. No change detected in normalized H3K27Me3 acid extracted histone levels between ASXL3^+/+ and ASXL3^+/fs fibroblasts, n = 3 (Error bars denote SD; ***P = 0.001).

Figure 4.

Transcriptome analysis of ASXL3^+/fsfibroblasts. (A) Percentage of differentially up- and down- regulated transcripts in ASXL3^+/fs fibroblasts. Five percent of differentially expressed transcripts were identified as non-coding RNA. (B) Volcano plot: all 564 significant DEGs (green dots) are represented in terms of their measured expression change (log₂ fold change) and the significance of their change [−log₁₀ (P-value)]. The significance is represented in terms of the negative log of the P-value. (C) Red circles are KEGG pathway with significant normalized enrichment scores according to the false discovery rate P-value (FDR P-value) corrected for multiple hypothesis testing. The impact of each pathway is plotted relative to the number of DEGs enriched in each pathway (p-Acc) and the perturbation (p-ORA) of the pathway based on the measured expression changes across the pathway topology. Black dots represent KEGG pathways that do not reach significance according to FDR P-value calculated. (D) Significantly enriched KEGG pathways and their FDR P-value.